Unitree CEO Says Humanoid Robots Could Reach a ChatGPT Moment in Two to Three Years
Key Takeaways
- •Wang Xingxing projects a ChatGPT-style breakthrough for humanoid robots in two to three years under ideal conditions, or five to ten years if progress slows.
- •He defines the industry tipping point as robots completing about 80% of tasks in 80% of unfamiliar environments, and cites residual error in decision-making models as the sector's biggest bottleneck.
- •Unitree became the first humanoid robot maker to list on China's A-share market, closing its debut session 460.34% above the 150.80 yuan offer price at a valuation of 340 billion yuan before falling sharply the next day.
- •Unitree shipped more than 5,500 humanoid robots in 2025, with most buyers still being universities and research institutions rather than factories.
- •China delivered more than 40,000 humanoid robots in the first half of this year, representing 97% of global shipments, while a State Council research center projects the physical AI market to exceed one trillion yuan by 2035.

Humanoid robots could reach a “ChatGPT moment” within two to three years under ideal conditions, or in five to ten years if progress slows, Unitree founder Wang Xingxing said at the World Robot Conference in Beijing on Thursday. His remarks came one day after Unitree’s Shanghai listing saw the stock surge more than 460% before later dipping.
Unitree CEO outlines an industry tipping point
Wang said he wants a robot that can enter a home or workplace it has never encountered before and complete about 80% of tasks using only spoken or typed instructions. He described the same threshold as being able to perform about 80% of tasks in 80% of unfamiliar environments, calling it “an important tipping point for the robot industry to usher in explosive growth.”
When OpenAI’s chatbot launched in late 2022, it quickly shifted large language models from a laboratory novelty to a product marketed at scale within weeks. Robots that can read and navigate physical space have not yet seen a comparable breakthrough, and Wang said the industry is still waiting for its own version of that leap.
For the sector, that comparison matters because the gap is not only about better software but about getting machines to behave reliably outside controlled demos. Wang also explained why robots fail in practice. He said robots trained for a period in a fixed room can come close to a perfect success rate, but they break down as soon as an object is replaced or the environment changes. He said the underlying problem remains the gap between digital models and physical hardware.
According to Wang, residual error is the biggest bottleneck facing AI embodied in physical robots, and decision-making models remain the sector’s central weakness.
He also acknowledged that Unitree is still far from fully applying physical AI in real-world settings, even though world models account for the company’s largest share of capital and staff.
Caution follows Unitree’s sharp stock move
Unitree became the first humanoid robot maker to list on China’s A-share market on Wednesday. Its shares closed that day 460.34% above the 150.80 yuan offer price, valuing the Hangzhou-based company at 340 billion yuan, or about $50.42 billion. The stock then fell sharply on Thursday.
“Relative to the mood around World Robot Conference and Unitree’s spectacular IPO, Wang Xingxing was notably sober about current capabilities,” Georg Stieler, head of automation at consultancy Stieler, told Technology.org.
Not everyone in the sector shares the same timeline. Wang He, founder of startup Galbot, said he expects the industry to reach its ChatGPT moment by 2028, defining that point as robots handling 70% to 80% of everyday tasks without special training.
Unitree’s prospectus shows that most of its buyers are still universities and research institutions rather than factories, underscoring how early the commercial market remains. The company said it shipped more than 5,500 humanoid robots in 2025.
China keeps expanding support for humanoid robots
China delivered more than 40,000 units in the first half of this year and accounted for 97% of global shipments, according to a Chinese industry body report cited by Technology.org.
A shrinking workforce is making machines more attractive for repetitive and dangerous jobs, even if they are still less efficient than human workers. A State Council research center has projected the physical AI market at 400 billion yuan by 2030 and more than one trillion yuan by 2035.
Beijing’s economic planners have urged manufacturers to stop flooding the market with nearly identical designs. Meanwhile, the U.S. Federal Communications Commission banned future imports of foreign-made humanoid and quadruped robots on national security grounds last month.